Michael Bacina | ⚖️ | 📍Cayman

@GivnerAriel You’re still well early.

I’m a bit late to the game here, but this just changed my life.

Well it's more a proposal but it's definitely and interesting one. Not sure if the economics work given wear and tear on a a mirror in space on a satellite / fuel etc but it's still a cool idea.

@gagerajusalicki @austincampbell A plaintiff lawyer seeking ex parte freezing orders is unlikely to have in mind keeping a freezing order narrow so as not to catch innocent people - their incentive is to capture the target assets and if it catches some extra well that can be adjusted.

@austincampbell lawyer here who has seen some absolutely insane court orders freezing crypto wallets. austin is correct, judges will sometimes just issue a TRO or PI and then you have to, well, just deal with it speaks to the crucial importance of explaining the technology extremely well

My guy, this is working exactly as anyone who has dealt with the US judiciary in traditional cases would expect. Federal judges are largely not tech experts. Likewise with commingled funds, you will get frozen first and then let people out quickly if they are not involved. Not unique to crypto! Go look at all the clawbacks and allocation in the Madoff case. Exactly this is how it works. This means the entire crypto ecosystem putting USD stablecoins with freeze and seize capabilities in commingled pools where the only option for the issuer is to freeze it all are sitting on many ticking time bombs. There can and will be freeze orders. Those are not final orders or seizures but the whole purpose is to lock up funds until you know more, and no US issuer is going to be in a position to tell a federal court to go pound sand if they want to keep their licenses and stay out of jail. This should not be surprising and it is not illegal. This has been a foreseeable problem for many years that very few protocols have bothered to consider and very few chains have bothered to build governance for. However, if you want to use US dollars, you are going to find that the strings attached are real, and that they will find you. If you don’t want to build for this reality or dislike that it is true, you can just use BTC or ETH! This isn’t possible with those tokens (nor should it be), but if you want the dollars part, you are also taking on dollar norms, which one hundred percent is going to include the exact same kinds of things that happen in traditional banking. To blame Circle for obeying a court order is… odd. Do you think they should have ignored the court? To blame the judge or the litigants is premature until we see what is sealed. But also I will remind everyone this is not a final order, just the start, and this is how litigation works… Crypto needs to start building for this and understand you cannot mix permissionless systems, RWA, and complex layered smart contracts or commingled wallets in a post-Genius world.

@BitGrateful For some reason when I read this post my first thought was: “It trains the model or it gets the hose again.”

Lawyers, should we update our email signatures to remind clients not to feed our communications into free AI models? It trains the model, so surely it waives privilege?

@jabranthelawyer Qui audet adipiscitur.

In 2013, I was told I was crazy to leave pharmacy for law I was nearly top of my class. I had potential to buy into a pharmacy business. A real career, a clear path and people telling me I'd made it I left anyway. In 2021, I was told I was crazy again when I decided to leave to open my own law firm I was the youngest senior associate in the firm's history. I had an 18 month old daughter and limited savings I left anyway. Taking risks for yourself is the only way to build a life you're actually happy with. Drown out the noise. Get comfortable being uncomfortable It's not easy. But then again, who said life was supposed to be? Grateful to be crazy.

@MetaLawMan Which traditional agency law might that be? As AI agents don't have their own legal personality, there would be no apparent basis for them to be treated as agents at law. They are just the user’s software actions, so wouldn’t liability be a straight line to the user?

Before you launch your AI AGENT into the world . . . Remember, under traditional Agency Law, you (as principal) can be liable for any damage caused by your AI agent. That liability can be to a counterparty to a transaction or to any third party harmed by actions of your agent. Be careful out there. (Not legal advice. I would encourage you to consult a real live attorney before launching AI Agents into the wild.)

@BitGrateful @AnthropicAI Will it starts suggesting “hey this doesn’t sound too urgent, can I run it off-peak”?

I heard from someone that @AnthropicAI is scrambling to reduce our usage, while adding features to encourage us to use it at the same time. The 2x off-peak discount isn't meant to get you hooked. it's meant to move you to off-peak... and it isn't working. Something about this model feels unsustainable.

What a world, where the unregulated Tether might just be delivering payment services to the developing world, a long stated goal of crypto.

The Last Chapter in the Book of Howey? SEC and CFTC Draw the Lines on Crypto linkedin.com/pulse/last-cha… via @LinkedIn

@BitGrateful We are heading to a world where IT tech support is no longer a mechanic fixing problems, but a mechanic guiding AI to build your next toolset and features. I cannot wait to ditch “good enough” SaaS tools which fall annoyingly slightly short of true productivity.

Business idea: one app that does everything

@Shaun__Furman @BitGrateful Retaining transcripts and being able to query them is *incredibly* useful - inside a privileged context - doing it in a non-privileged world is just a liability risk.

@BitGrateful At most I write 5-10 bullet points from a call. What is going on in these calls that everyone needs every word recorded all of a sudden?

Would you transcribe every call, every meeting, every conversation into claude, so you never miss or forget anything ever again? Or is that too much…?

@BitGrateful I had been sketching out the same thing, initially wanting to fix the frustratingly painful parts of our matter management SaaS software but increasingly seeing the raw value and efficiency in this.

Slowly building myself a CRM, adding my clients and matters to it, integrating the tools I use, all proprietary and vibe coded and local. This is way better for me than the expensive cloud guys.

More lawyers in trouble for AI misuse today: Attorney represented a criminal defendant and filed an application to reopen an appeal. The application alleged prosecutorial misconduct and cited quotes attributed to the prosecutor. Only problem was the quotes were completely made up. "The page Appellant cites to on page 4 for the 'legally inflammatory' statement by the prosecutor- is, in fact, the court reporter’s signature page, with no statements of any type by the prosecutor." The fake quotes came from ChatGPT. A paralegal uploaded case materials to ChatGPT and pasted the output into the application. The lawyer didn't catch it. But it gets worse: after the application was denied, the lawyer appealed the denial to the Ohio Supreme Court anyway. And when he submitted his firms "AI Policy" to show he was taking corrective action, the court determined that the policy itself was AI-generated and was incomplete. "The proffering of an AI-generated AI policy as a remedial measure in a case involving the submission of AI-generated fabrications to this court is, at best, ironic." And two months after the sanctions hearing in this case, the attorney did it again, in another case. He submitted a filing with the ChatGPT prompt embedded in the filing itself: "Would you like me to draft the next argument section (e.g., argument 1 – B on the 'nature of the charge' omission) in the same tone and format so your brief reads as a seamless multi-print memorandum?" Sanctions: -$2,000 fine -Referral to Ohio Office of Disciplinary Counsel -Must serve copy of judgment on judge of every court in which he makes an appearance, for 2 years -Must include certification that all cases are real and verified, for 2 years -6 hours mandatory CLE about AI ethics -Must write apology letters to prosecutor, trial judge, trial defense counsel, and prior appellate counsel who were defamed by fake quotes.

Can we ship in Australian journalists? Because of course this was one of them. Their entire country is better at our national media than we are.

@jabranthelawyer 100%

@MikeBacina Thanks ser! The AI law offerings are pretty awful and way too overpriced. One of my goals is to truly disrupt that market. Definitely a much more fun time to be a lawyer :)

I've spent the last few weeks rebuilding how I work with AI tools Narrowed it down to Claude and NotebookLM across real client work Honestly, didn't expect them be this useful this fast OpenClaw joins the mix in April. Better late than never. But wanted to reduced my client facing commitments, so I could give it a proper go The question is going to be whether the combination of tools produces meaningfully better outputs Going to be a fun experiment, so I'll be documenting what actually works and what doesn't

@jabranthelawyer Nice write up! I would add a fourth: we overestimate the technological impact in the short term but underestimate it in the long term.

The debate about AI and work keeps producing the same two camps "AI destroys jobs" vs "Everything will be fine" Both are right. But both are incomplete. There's three economic theories that together give us clearer answers that: 1. the pie does get bigger 2. the gains won't distribute quickly or evenly, and 3. for specific workers in specific roles, the transition produces real structural casualties regardless of what happens at the macro level These aren't contradictions They're three forces operating simultaneously across every major technological transition for the past 200 years

One of the recurring mistakes in technology policy is to assume that every new tool requires a new body of law. But a recent Nebraska amicus brief gets it exactly right. In a case involving hallucinated case citations in court filings, the Nebraska AG and SG argue against AI-specific ethics or certification rules. I agree that the better view is that the existing rules of professional conduct already do the work. Lawyers have long owed duties of competence and diligence. If a lawyer files fabricated authority—whether because of sloppiness, copy-paste errors, or blind reliance on an LLM—the problem doesn't have much to do with AI. It's that the lawyer has violated ordinary professional obligations. Nebraska's DOJ deserves credit for taking the right approach here. There is always temptation to grandstand about AI, to stoke fear, or to pile on special disclosure and certification requirements that make practice more cumbersome while adding little of substance. But rules aimed specifically at “AI misuse” often just duplicate duties lawyers already have, while increasing complexity, compliance costs, and the risk that useful innovation gets chilled. That is a healthy model for legal adaptation to new technology: enforce general principles faithfully, resist bespoke panic, and add new law only when old law truly fails. (Note that the Fourth Circuit’s recent public admonishment points in the same direction, In re Erich Chibueze Nwaubani. The court sanctioned a lawyer who cited nonexistent cases, applying the rules already on the books to do so.) Thanks to Eric Goldman for flagging the Fourth Circuit case, and to Zachary Pohlman for sharing his brief to the Nebraska Supreme Court. He's #3 on the brief, which, for the benefit of my law students, means he did A LOT of the heavy lifting on this! #LegalEthics #AI #TechnologyPolicy

Run your own genomic test locally with AI to help live a healthier life? 🤯

This is 100% going to happen to people who aren’t using real lawyers, an arms race on prompt attacks.